Or, how to go from nothing to something!
I get a lot of questions on growth, what the process of growth is, and how as a growth team you approach problems thought a growth mindset. And so I’ll walk you through the step functions of growth as best as I can. As a primer, growth teams typically will work with a hypothesis about some portion of the marketing funnel or product, and build a test to validate or invalidate the hypothesis.
Example hypothesis
You may start with something like this: Because shoppers to my site are adding items to their cart but not checking out right away, by emailing them a couple hours later reminding them to come back to their cart, I’ll increase the conversion rate of the people that entered the funnel in the first place by doing so.
Typically, this type of test in the e-commerce space would be called an abandoned cart sequence. However, building something like an abandoned cart in it’s fully baked form may take a lot of resources. So the way the step functions in growth work are that you go from 0 to 1, 1 to 10, 10 to 100, and then what I call “building a program,” which is the last step when you scale your efforts. It sort of works like a set of steps (hence the name).
The 0 to 1 step
And so the step function between 0 and 1 means that you’re simply going from not doing something to doing something. In the case of abandoned cart, you may not even have been aware that you had a significant number of people adding items to their cart and then leaving. So your step function there is to just get in the mindset of understanding what the key metrics are, and what the user behavior is, and make a decision that you want to try to validate this hypothesis that sending an email to those people is going to move the needle.
In the early days of the experiment what you likely do is the following:
- Manually pick a small cohort of people who add items to the cart
- Run a query on the database where you have your funnel information stored to figure out who those people are (this is assuming that you’ve got their email address already, so they’re probably either existing customers or you’ve captured their email along the way)
- You write a simple script, and manually fire off 100 emails over a week of time to those users, and you track the results.
- You determine directionally if the test was a success or failure
That’s really the gist 0 to 1 test, where you went from doing nothing to now doing something. So in your 0 to 1 step function series, you may have a half a dozen test ideas that you want to run. Now, a lot of tests will end in the 0 to 1 area. Perhaps your hypothesis was proven wrong. Maybe none of the 100 ended up checking out. Perhaps you even were told to stop emailing some folks. Or, maybe directionally you think you are onto something. The numbers don’t quite make a ton of sense yet, but you aren’t getting screamed at by your customers, and you didn’t bet the company on this test. So then you have to decide, “Do I want to invest a little bit more time and run some additional experiments within my hypothesis or not?” It’s worth noting that at this point, you probably don’t have statistical significance yet on the test results, but that’s ok. Keep going and listen to your gut in this case.
The 1 to 10 step
Then, your next step function might be a 1 to 10. And this is where you’ll probably get a little bit more scientific with your test. If you’ve got a data science team, you may pull them in here. If you’ve got a design team, maybe you’ll pull them in to make the emails look better. If you’ve got a messaging team, maybe you’ll have somebody write better content than what you had first used.
The 0 to 1 game is meant to be very lo-fi, maybe even kind of growth hacky. It’s likely very, very low risk, kind of messy, and certainly not perfect. In the 1 to 10 realm, you’re starting to bake the idea a little bit more, so you’re getting a little bit more polished. And then you measure that, and you may have iterations of testing happening in that stage. Let’s say that you see a 5% or a 10% increase in conversions from the initial 0 to 1 test. In the 1 to 10 phase, you might say, “Well, what happens if I change the subject line or if I change the copy? Can I get a 15% improvement?” Then you run that. Again, you look at your stats, hear what your customers are saying, and evaluate if this is worth expanding even more.
The 10 to 100 step
Here we are at a 10 to 100 step function. With your 10 to 100 test – depending on how big your team is and how many users you have – you increase the visibility of the test by 25% or 50%, or even point 50-75% of your total user base to this variant if you’re still early and don’t have a large number of users. If you fail at that point, then you probably have statistical significance, and maybe it was not a great idea. Maybe it’s a failed test. But If it does move the needle, then that might be a test that becomes a feature. In this case you likely already have some buy-in from product and engineering to build the feature, and at this point you are likely ready to pitch to get your feature implemented. For our example, at this point we’d be looking at building out the whole abandoned cart life cycle within the product. And from here, we optimize and scale the feature once it’s live. Your little 0 to 1 test just becomes part of your software, and then you move on to other tests.
So hopefully, that’s helpful. Those are the step functions of growth!
If you’d like to see an example of a good test turned program, check out some work my growth team and I did at Upside Travel a while back: https://engineering.upside.com/testing-in-the-trenches-how-generic-vs-personalized-content-impacts-conversion-rates-468208445a11
